redux-loop

A port of the Elm Architecture to Redux that allows you to sequence your effects naturally and purely by returning them from your reducers.

Isn't it incorrect to cause side-effects in a reducer?

Yes! Absolutely.

Doesn't redux-loop put side-effects in the reducer?

It doesn't. The values returned from the reducer when scheduling an effect with redux-loop only describe the effect. Calling the reducer will not cause the effect to run. The value returned by the reducer is just an object that the store knows how to interpret when it is enhanced by redux-loop. You can safely call a reducer in your tests without worrying about waiting for effects to finish and what they will do to your environment.

What are the environment requirements for redux-loop?

redux-loop requires polyfills for ES6 Promise and Symbol to be included if the browsers you target don't natively support them.

Why use this?

Having used and followed the progression of Redux and the Elm Architecture, and after trying other effect patterns for Redux, we came to the following conclusion:

Synchronous state transitions caused by returning a new state from the reducer in response to an action are just one of all possible effects an action can have on application state.

Many other methods for handling effects in Redux, especially those implemented with action-creators, incorrectly teach the user that asynchronous effects are fundamentally different from synchronous state transitions. This separation encourages divergent and increasingly specific means of processing particular types effects. Instead, we should focus on making our reducers powerful enough to handle asynchronous effects as well as synchronous state transitions. With redux-loop, the reducer doesn't just decide what happens now due to a particular action, it decides what happens next. All of the behavior of your application can be traced through one place, and that behavior can be easily broken apart and composed back together. This is one of the most powerful features of the Elm architecture, and with redux-loop it is a feature of Redux as well.

Tutorial

Install the store enhancer

import { createStore, compose, applyMiddleware } from 'redux';
import reducer from './reducers';
import { install } from 'redux-loop';
import someMiddleware from 'some-middleware';

const enhancer = compose(
  applyMiddleware(someMiddleware),
  install()
);

const store = createStore(reducer, initialState, enhancer);

Installing redux-loop is as easy as installing any other store enhancer. You can apply it directly over createStore or compose it with other enhancers and middlewares. Composition of enhancers can be confusing, so the order in which install() is applied may matter. If something like applyMiddleware() doesn't work when called before install(), applying after may fix the issue.

Write a reducer with some Cmds (side effects)

import { loop, Cmd } from 'redux-loop';

function initAction(){
    return {
      type: 'INIT'
    };
}

function fetchUser(userId){
    return fetch(`/api/users/${userId}`);
}

function userFetchSuccessfulAction(user){
   return {
      type: 'USER_FETCH_SUCCESSFUL',
      user
   };
}

function userFetchFailedAction(err){
   return {
      type: 'USER_FETCH_ERROR',
      err
   };
}

const initialState = {
  initStarted: false,
  user: null,
  error: null
};

function reducer(state = initialState, action) {
  switch(action.type) {
  case 'INIT':
    return loop(
      {...state, initStarted: true},
      Cmd.run(fetchUser, {
        successActionCreator: userFetchSuccessfulAction,
        faiLActionCreator: userFetchFailedAction,
        args: ['123']
      })
    );

  case 'USER_FETCH_SUCCESSFUL':
    return {...state, user: action.user};
    
  case 'USER_FETCH_FAILED':
    return {...state, error: action.error};
    
  default:
    return state;
  }
}

Any reducer case can return a loop instead of a state object. A loop joins an updated model state with a cmd for the store to process. A cmd is just an object that describes what side effect should run, and what to do with the result. There are several options for cmds, all available under the Cmd object:

• run(functionToCall, options)
  • Accepts a function to run and options for how to call it and what to do with the result.
• action(actionToDispatch)
  • Accepts an actionToDispatch instance to dispatch immediately once the current dispatch cycle is completed.
• batch(cmds)
  • Accepts an array of other cmds and runs them in parallel, dispatching the resulting actions in their original order once all cmds are resolved.
• sequence(cmds)
  • The same as batch, but commands wait for the previous command to finish before starting.
• none()
  • A no-op action, for convenience.

Accessing state and dispatching actions from your Cmds

While most of your side effect methods should be agnostic to redux, sometimes you will need the flexibility of accessing arbitrary slices of state or dispatching an action separately from the result of the cmd. Here's an example of how you can do this.

import {loop, Cmd} from 'redux-loop';
import {doSomething} from 'something.js';
import {doSomethingResultAction} from './actions.js';
function reducer(state, action) {
  switch(action.type) {
  case 'ACTION':
    return loop(
      {...state, initStarted: true},
      Cmd.run(doSomething, {
         successActionCreator: doSomethingResultAction,
         args: [Cmd.getState, Cmd.dispatch]
      })
    );
  default:
    return state;
  }
}

//something.js
export function doSomething(getState, dispatch){
   let value = getState().some.random.value;
   dispatch(someRandomAction());
}

Cmd.getState and Cmd.dispatch are just symbols that the enhanced store will replace with the actual functions at the time the cmd is executed. You can use these symbols as parameters to your cmds as you would any other parameter (the order does not matter).

Easily test reducer results

import test from 'tape';
import reducer, { fetchDetails } from './reducer';
import { loadingStart, loadingSuccess, loadingError } from './actions';
import { Effects, loop } from 'redux-loop';

test('reducer works as expected', (t) => {
  const state = { loading: false };

  const result = reducer(state, loadingStart(1));

  t.deepEqual(result, loop(
    { loading: true },
    Cmd.run(fetchDetails, {
      sucessActionCreator: loadingSuccess,
      failActionCreator: loadingError,
      args: [1]
    })
  ));
});

Cmds are declarative specifications of the next behavior of the store. They are only processed by an active store, pushing effecting behavior to the edge of the application. You can call a reducer as many times with a given action and state and always get a result which is deepEqual.

CAVEAT For testing sanity, always pass a referenceable function to a cmd. Functions curried or bound from the same function with the same arguments are not equal within JavaScript, and so are best to avoid if you want to compare effects in your tests.

combineReducers with redux-loop

import { createStore } from 'redux';
import { combineReducers, install } from 'redux-loop';

import { firstReducer, secondReducer } from './reducers';

const reducer = combineReducers({
  first: firstReducer,
  second: secondReducer,
});

const store = createStore(reducer, initialState, install());

The combineReducers implementation in redux-loop is aware that some of your reducers might return cmds, and it knows how to properly compose them and forward them to the store.

How to use combineReducers when using libraries such as immutable.js

Our combineReducers can also handle states made of data structures other than the default {}, you simply pass it in the root state, an accessor function (which returns a value for that key), and a mutator function (which returns a new version of the object with a value set at a given key). The example below demonstrates using Immutable.js' Map() data structure, but you can use any key => value data structure as long as you provide your own accessor and mutator functions.

While these customization features are provided for convenience, if you are using using something other than a plain object as your state, you should consider writing your own implementation of combineReducers, using our version as a reference. It's impossible to provide a generic combineReducers that will be optimal for all state shapes. As an example, updating an immutable.js state would be much faster if you made use of the withMutations method to batch updates.

import { combineReducers } from 'redux-loop';
import { Map } from 'immutable';

import { firstReducer, secondReducer } from './reducers';

const reducers = {
  first: firstReducer,
  second: secondReducer,
}

//Map() is now used as the new root state, and custom accessor and mutator properties are provided
const reducer = combineReducers(
    reducers,
    Map(),
    (child, key) => child.get(key),
    (child, key, value) => child.set(key, value)
);

Avoid circular loops!

function reducer(state, action) {
  switch (action.type) {
    case 'FIRST':
      return loop(
        state,
        Cmd.action(second())
      );

    case 'SECOND':
      return loop(
        state,
        Cmd.action(first())
      );
  }
}

This minimal example will cause perpetual dispatching! While it is also possible to make this mistake with large, complicated networks of redux-thunk action creators, it is much easier to spot the mistake before it is made. It helps to keep your reducers small and focused, and use combineReducers or manually compose reducers so that the number of actions you deal with at one time is small. A small set of actions which initiate a loop will help reduce the likelihood of causing circular dispatches.

Support

Potential bugs, generally discussion, and proposals or RFCs should be submitted as issues to this repo, we'll do our best to address them quickly. We use this library as well and want it to be the best it can! For questions about using the library, submit questions on StackOverflow with the redux-loop tag.

Don't see a feature you want?

If you're interested in adding something to redux-loop but don't want to wait for us to incorporate the idea you can follow these steps to get your own installable version of redux-loop with your feature included:

1. Fork the main repo here
2. Add your feature or change
3. Change the package "name" in package.json to be "@<your-npm-username>/redux-loop
4. Commit to master and npm publish
5. npm install @<your-npm-username>/redux-loop

We are always interested in new ideas, but sometimes we get a little busy and fall behind on responding and reviewing PRs. Hopefully this process will allow you to continue making progress on your projects and also provide us with more context if and when you do decide to make a PR for your new feature or change. The best way to verify new features for a library is to use them in real-world scenarios!

Contributing

Please note that this project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms. Multiple language translations are available at contributor-covenant.org